Date:    Monday March 8, 4:30pm
Speaker: Amit Chakrabarti

TITLE: Randomized Rounding of Semidefinite Relaxations

ABSTRACT: Developing approximation algorithms for NP-complete
problems is an important area of theoretical computer science.
Recent work has shown that for some important problems, solving
a semidefinite programming relaxation followed by "rouding" the
solution to an integer solution using a randomized rounding
scheme, provides the best known approximation factors.

In this talk, I shall describe algorithms for MAX-CUT, MAX-2SAT,
MAX-SAT and GRAPH-COLORING which use this approach. No prior
exposure to the theory of semidefinite programming will be
required.